NK_Killing_Mixed_Model
Akshay Iyer
2024-04-18
Study Design and Aims
NK cells from HIV Exposed Infected (HEI) and HIV Exposed Uninfected (HEU) from the TARA cohort at Entry (1-2 months of age) and at 12 months of age were tested for functionality/ability to kill. The treatments they were exposed to were;
1.CEM 2.CEM+IL15 3.HUT78 4.HUT78+IL15 5.K562 6.K562+IL15
To deal with the paired nature of the data, at different timepoints, a mixed effects model was used to analyse the data, correcting for batch effect. It was run over all the data and also for each treatment seperately
All Treatments
Heatmap
Mixed Effects Model Summary
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: `specific killing` ~ (HIV + Timepoint + gender + Treatment)^2 +
## (1 | PID)
## Data: TARA_Killing
##
## REML criterion at convergence: 1401
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.6756 -0.5255 -0.0768 0.4108 3.4438
##
## Random effects:
## Groups Name Variance Std.Dev.
## PID (Intercept) 18.28 4.275
## Residual 32.42 5.694
## Number of obs: 232, groups: PID, 25
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 4.4923 2.1800 68.7873 2.061 0.0431
## HIVHEU -0.9731 3.3657 39.9031 -0.289 0.7740
## Timepoint12 -0.3546 1.9752 185.8241 -0.179 0.8577
## gendermale -2.9967 2.9554 47.3507 -1.014 0.3157
## TreatmentCEM+IL15 -0.4982 2.8045 186.2790 -0.178 0.8592
## TreatmentHUT78 8.7915 2.1767 183.6368 4.039 7.88e-05
## TreatmentHUT78+IL15 13.0393 2.2990 184.7132 5.672 5.38e-08
## TreatmentK562 12.4722 2.1767 183.6368 5.730 4.05e-08
## TreatmentK562+IL15 23.7723 2.2495 184.3933 10.568 < 2e-16
## HIVHEU:Timepoint12 2.2508 1.6836 192.2791 1.337 0.1828
## HIVHEU:gendermale -0.1113 3.9588 20.4402 -0.028 0.9778
## HIVHEU:TreatmentCEM+IL15 2.8282 3.2017 184.9005 0.883 0.3782
## HIVHEU:TreatmentHUT78 0.6330 2.5763 184.0253 0.246 0.8062
## HIVHEU:TreatmentHUT78+IL15 3.6824 2.7663 184.5462 1.331 0.1848
## HIVHEU:TreatmentK562 1.3791 2.5763 184.0253 0.535 0.5931
## HIVHEU:TreatmentK562+IL15 1.5682 2.6449 183.7191 0.593 0.5540
## Timepoint12:gendermale 3.6164 1.6714 194.8585 2.164 0.0317
## Timepoint12:TreatmentCEM+IL15 0.6095 3.1317 185.9417 0.195 0.8459
## Timepoint12:TreatmentHUT78 0.4754 2.4708 183.9292 0.192 0.8476
## Timepoint12:TreatmentHUT78+IL15 -1.9288 2.6345 183.9783 -0.732 0.4650
## Timepoint12:TreatmentK562 0.7428 2.4708 183.9292 0.301 0.7640
## Timepoint12:TreatmentK562+IL15 0.6144 2.5463 183.4599 0.241 0.8096
## gendermale:TreatmentCEM+IL15 2.4896 3.1145 185.2528 0.799 0.4251
## gendermale:TreatmentHUT78 -2.6745 2.5114 183.9501 -1.065 0.2883
## gendermale:TreatmentHUT78+IL15 -1.6134 2.6608 184.3140 -0.606 0.5450
## gendermale:TreatmentK562 -4.1412 2.5114 183.9501 -1.649 0.1009
## gendermale:TreatmentK562+IL15 -5.4460 2.5842 183.8411 -2.107 0.0364
##
## (Intercept) *
## HIVHEU
## Timepoint12
## gendermale
## TreatmentCEM+IL15
## TreatmentHUT78 ***
## TreatmentHUT78+IL15 ***
## TreatmentK562 ***
## TreatmentK562+IL15 ***
## HIVHEU:Timepoint12
## HIVHEU:gendermale
## HIVHEU:TreatmentCEM+IL15
## HIVHEU:TreatmentHUT78
## HIVHEU:TreatmentHUT78+IL15
## HIVHEU:TreatmentK562
## HIVHEU:TreatmentK562+IL15
## Timepoint12:gendermale *
## Timepoint12:TreatmentCEM+IL15
## Timepoint12:TreatmentHUT78
## Timepoint12:TreatmentHUT78+IL15
## Timepoint12:TreatmentK562
## Timepoint12:TreatmentK562+IL15
## gendermale:TreatmentCEM+IL15
## gendermale:TreatmentHUT78
## gendermale:TreatmentHUT78+IL15
## gendermale:TreatmentK562
## gendermale:TreatmentK562+IL15 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Specific Killing
Specific Killing
Predicted Killing
Estimated Means Pairwise Comparison
HIV, Treatment: These indicate the subgroup for which the comparison is made. For example, “HIV = HEU, Treatment = CEM” refers to participants who are HIV positive under the CEM treatment.
Timepoint_pairwise: This specifies the comparison being made, here between “Entry” and “12” timepoints.
Non-Pairwise EMMs Interpretation: Positive EMM at “12”: Indicates that the adjusted mean value of your outcome variable (e.g., specific killing) is positive at the 12-month timepoint. This value is an estimate that accounts for other factors in your model. Negative EMM at “Entry”: Suggests that the adjusted mean value of your outcome variable at the entry timepoint is negative. Again, this is an adjusted estimate considering the model’s covariates.
Pairwise Comparisons Interpretation: Negative Estimate for “Entry - 12” Comparison: In pairwise comparisons, a negative estimate indicates that the mean value of the outcome variable at the first timepoint (“Entry”) is lower than at the second timepoint (“12”). Essentially, this tells you there’s an increase from “Entry” to “12”. The interpretation of the sign here depends on how the comparison is framed.
SE (Standard Error): Indicates the precision of the estimate; a smaller SE suggests a more precise estimate.
df (Degrees of Freedom): Reflects the sample size and complexity of the model; more degrees of freedom generally indicate a larger sample size or less complex model.
t.ratio: The ratio of the estimate to its standard error. A larger absolute value of the t-ratio indicates a stronger evidence against the null hypothesis (no difference).
p.value: The probability of observing the data (or something more extreme) if the null hypothesis were true. A small p-value (e.g., < 0.05) suggests that the observed difference is unlikely to have occurred by chance.
Significance: A summary of the p-value, often marked with asterisks to indicate levels of significance (e.g., “***” for p < 0.001).
Timepoint
Simple
Summary
## HIV = HEI, Treatment = CEM:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect -0.7268221 0.9661682 187.94 -0.752 0.4528 ns
## 12 effect 0.7268221 0.9661682 187.94 0.752 0.4528 ns
##
## HIV = HEU, Treatment = CEM:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect -1.8522167 1.0419269 188.37 -1.778 0.0771 ns
## 12 effect 1.8522167 1.0419269 188.37 1.778 0.0771 ns
##
## HIV = HEI, Treatment = CEM+IL15:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect -1.0315967 1.3225490 189.40 -0.780 0.4364 ns
## 12 effect 1.0315967 1.3225490 189.40 0.780 0.4364 ns
##
## HIV = HEU, Treatment = CEM+IL15:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect -2.1569912 1.4320696 190.51 -1.506 0.1337 ns
## 12 effect 2.1569912 1.4320696 190.51 1.506 0.1337 ns
##
## HIV = HEI, Treatment = HUT78:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect -0.9645334 0.9301835 187.52 -1.037 0.3011 ns
## 12 effect 0.9645334 0.9301835 187.52 1.037 0.3011 ns
##
## HIV = HEU, Treatment = HUT78:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect -2.0899279 0.9980884 186.99 -2.094 0.0376 *
## 12 effect 2.0899279 0.9980884 186.99 2.094 0.0376 *
##
## HIV = HEI, Treatment = HUT78+IL15:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect 0.2375769 1.0355580 188.19 0.229 0.8188 ns
## 12 effect -0.2375769 1.0355580 188.19 -0.229 0.8188 ns
##
## HIV = HEU, Treatment = HUT78+IL15:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect -0.8878176 1.1159195 187.53 -0.796 0.4273 ns
## 12 effect 0.8878176 1.1159195 187.53 0.796 0.4273 ns
##
## HIV = HEI, Treatment = K562:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect -1.0981978 0.9301835 187.52 -1.181 0.2392 ns
## 12 effect 1.0981978 0.9301835 187.52 1.181 0.2392 ns
##
## HIV = HEU, Treatment = K562:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect -2.2235923 0.9980884 186.99 -2.228 0.0271 *
## 12 effect 2.2235923 0.9980884 186.99 2.228 0.0271 *
##
## HIV = HEI, Treatment = K562+IL15:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect -1.0340152 0.9911364 187.95 -1.043 0.2982 ns
## 12 effect 1.0340152 0.9911364 187.95 1.043 0.2982 ns
##
## HIV = HEU, Treatment = K562+IL15:
## contrast estimate SE df t.ratio p.value Significance
## Entry effect -2.1594098 1.0543409 188.24 -2.048 0.0419 *
## 12 effect 2.1594098 1.0543409 188.24 2.048 0.0419 *
##
## Results are averaged over the levels of: gender
## Degrees-of-freedom method: kenward-roger
## P value adjustment: fdr method for 2 testsPlot
Pairwise Timepoint
Summary
## HIV = HEI, Treatment = CEM:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 -1.453644 1.932336 187.94 -0.752 0.4528 ns
##
## HIV = HEU, Treatment = CEM:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 -3.704433 2.083854 188.37 -1.778 0.0771 ns
##
## HIV = HEI, Treatment = CEM+IL15:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 -2.063193 2.645098 189.40 -0.780 0.4364 ns
##
## HIV = HEU, Treatment = CEM+IL15:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 -4.313982 2.864139 190.51 -1.506 0.1337 ns
##
## HIV = HEI, Treatment = HUT78:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 -1.929067 1.860367 187.52 -1.037 0.3011 ns
##
## HIV = HEU, Treatment = HUT78:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 -4.179856 1.996177 186.99 -2.094 0.0376 *
##
## HIV = HEI, Treatment = HUT78+IL15:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 0.475154 2.071116 188.19 0.229 0.8188 ns
##
## HIV = HEU, Treatment = HUT78+IL15:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 -1.775635 2.231839 187.53 -0.796 0.4273 ns
##
## HIV = HEI, Treatment = K562:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 -2.196396 1.860367 187.52 -1.181 0.2392 ns
##
## HIV = HEU, Treatment = K562:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 -4.447185 1.996177 186.99 -2.228 0.0271 *
##
## HIV = HEI, Treatment = K562+IL15:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 -2.068030 1.982273 187.95 -1.043 0.2982 ns
##
## HIV = HEU, Treatment = K562+IL15:
## Timepoint_pairwise estimate SE df t.ratio p.value Significance
## Entry - 12 -4.318820 2.108682 188.24 -2.048 0.0419 *
##
## Results are averaged over the levels of: gender
## Degrees-of-freedom method: kenward-rogerPlot
HIV Status
Simple
Summary
## Timepoint = Entry, Treatment = CEM:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect 0.5143628 1.348080 66.41 0.382 0.7040 ns
## HEU effect -0.5143628 1.348080 66.41 -0.382 0.7040 ns
##
## Timepoint = 12, Treatment = CEM:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect -0.6110318 1.388833 72.48 -0.440 0.6613 ns
## HEU effect 0.6110318 1.388833 72.48 0.440 0.6613 ns
##
## Timepoint = Entry, Treatment = CEM+IL15:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect -0.8997529 1.621481 113.11 -0.555 0.5801 ns
## HEU effect 0.8997529 1.621481 113.11 0.555 0.5801 ns
##
## Timepoint = 12, Treatment = CEM+IL15:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect -2.0251475 1.661654 118.03 -1.219 0.2254 ns
## HEU effect 2.0251475 1.661654 118.03 1.219 0.2254 ns
##
## Timepoint = Entry, Treatment = HUT78:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect 0.1978470 1.331246 63.47 0.149 0.8823 ns
## HEU effect -0.1978470 1.331246 63.47 -0.149 0.8823 ns
##
## Timepoint = 12, Treatment = HUT78:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect -0.9275475 1.338343 64.72 -0.693 0.4908 ns
## HEU effect 0.9275475 1.338343 64.72 0.693 0.4908 ns
##
## Timepoint = Entry, Treatment = HUT78+IL15:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect -1.3268369 1.420246 77.92 -0.934 0.3531 ns
## HEU effect 1.3268369 1.420246 77.92 0.934 0.3531 ns
##
## Timepoint = 12, Treatment = HUT78+IL15:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect -2.4522314 1.457938 82.56 -1.682 0.0964 ns
## HEU effect 2.4522314 1.457938 82.56 1.682 0.0964 ns
##
## Timepoint = Entry, Treatment = K562:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect -0.1751853 1.331246 63.47 -0.132 0.8957 ns
## HEU effect 0.1751853 1.331246 63.47 0.132 0.8957 ns
##
## Timepoint = 12, Treatment = K562:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect -1.3005798 1.338343 64.72 -0.972 0.3348 ns
## HEU effect 1.3005798 1.338343 64.72 0.972 0.3348 ns
##
## Timepoint = Entry, Treatment = K562+IL15:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect -0.2697130 1.355781 67.58 -0.199 0.8429 ns
## HEU effect 0.2697130 1.355781 67.58 0.199 0.8429 ns
##
## Timepoint = 12, Treatment = K562+IL15:
## contrast estimate SE df t.ratio p.value Significance
## HEI effect -1.3951075 1.396551 73.63 -0.999 0.3211 ns
## HEU effect 1.3951075 1.396551 73.63 0.999 0.3211 ns
##
## Results are averaged over the levels of: gender
## Degrees-of-freedom method: kenward-roger
## P value adjustment: fdr method for 2 testsPlot
Pairwise Timepoint
Summary
## Timepoint = Entry, Treatment = CEM:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU 1.028726 2.696160 66.41 0.382 0.7040 ns
##
## Timepoint = 12, Treatment = CEM:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU -1.222064 2.777666 72.48 -0.440 0.6613 ns
##
## Timepoint = Entry, Treatment = CEM+IL15:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU -1.799506 3.242961 113.11 -0.555 0.5801 ns
##
## Timepoint = 12, Treatment = CEM+IL15:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU -4.050295 3.323307 118.03 -1.219 0.2254 ns
##
## Timepoint = Entry, Treatment = HUT78:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU 0.395694 2.662491 63.47 0.149 0.8823 ns
##
## Timepoint = 12, Treatment = HUT78:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU -1.855095 2.676686 64.72 -0.693 0.4908 ns
##
## Timepoint = Entry, Treatment = HUT78+IL15:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU -2.653674 2.840493 77.92 -0.934 0.3531 ns
##
## Timepoint = 12, Treatment = HUT78+IL15:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU -4.904463 2.915875 82.56 -1.682 0.0964 ns
##
## Timepoint = Entry, Treatment = K562:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU -0.350371 2.662491 63.47 -0.132 0.8957 ns
##
## Timepoint = 12, Treatment = K562:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU -2.601160 2.676686 64.72 -0.972 0.3348 ns
##
## Timepoint = Entry, Treatment = K562+IL15:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU -0.539426 2.711562 67.58 -0.199 0.8429 ns
##
## Timepoint = 12, Treatment = K562+IL15:
## HIV_pairwise estimate SE df t.ratio p.value Significance
## HEI - HEU -2.790215 2.793102 73.63 -0.999 0.3211 ns
##
## Results are averaged over the levels of: gender
## Degrees-of-freedom method: kenward-rogerPlot
Individual Treatments
This is the effect of different covariates on each NK subset, controlling for all other covariates
Covariate Analysis Untreated
HIV Effect
Timepoint Effect
Gender Effect
HUT78 Specific Killing Effect
K562 Specific Killing Effect
Covariate Analysis HUT78
HIV Effect
Timepoint Effect
Gender Effect
HUT78 Specific Killing Effect
Covariate Analysis K562
HIV Effect
Timepoint Effect
Gender Effect
K562 Specific Killing Effect